HDP on Linux – Installation Forum

/etc/hosts file

  • #33359
    Swapnil Patil

    I am confused with whether /etc/hosts file of all datanodes should contain IP of every datanode in the cluster?
    Do datanodes communicate with each other in the network? and if yes why?
    Hoping for the response
    thanks in advance

to create new topics or reply. | New User Registration

  • Author
  • #33371

    Yes! in order to send replication messages to each other,datanodes should know each other.

    As an example,when block #1 of data is going to be replicated on nodes 3,6,9 (data nodes on which block is going to be written are selected by NameNode), client machine sends the first node in the list of suggested datanodes(list provided by namenode) a message to inform that it is going to send a block of data.and also the list of other selected data nodes are transmitted via this message.the first datanode sends the 2nd datanode a similar message,and 2nd sends 3rd datanode(and so on… if more data nodes are selected) and the last datanode,sends back a message “I am ready” and then each data node sends back their previous datanode an “I am ready” message.when the first data node recieves that message,replication from client to first node starts.and when replication to first node completed,replication to 2nd starts from 1st node,and so on…

    You can read about the whole process here 😉

    Swapnil Patil

    @ Hesam
    Thanks alot for your reply..

You must be to reply to this topic. | Create Account

Support from the Experts

A HDP Support Subscription connects you experts with deep experience running Apache Hadoop in production, at-scale on the most demanding workloads.

Enterprise Support »

Become HDP Certified

Real world training designed by the core architects of Hadoop. Scenario-based training courses are available in-classroom or online from anywhere in the world

Training »

Hortonworks Data Platform
The Hortonworks Data Platform is a 100% open source distribution of Apache Hadoop that is truly enterprise grade having been built, tested and hardened with enterprise rigor.
Get started with Sandbox
Hortonworks Sandbox is a self-contained virtual machine with Apache Hadoop pre-configured alongside a set of hands-on, step-by-step Hadoop tutorials.
Modern Data Architecture
Tackle the challenges of big data. Hadoop integrates with existing EDW, RDBMS and MPP systems to deliver lower cost, higher capacity infrastructure.